Copyright ©ERS Journals Ltd 1998

Eur Respir J 1998; 11: 1392–1404 European Respiratory Journal

DOI: 10.1183/09031936.98.11061392 ISSN 0903 - 1936
Printed in UK - all rights reserved

REVIEW

Magnetic resonance (MR) imaging of the chest: state-of-the-art

R.C. Bittner, R. Felix

aa
Magnetic resonance (MR) imaging of the chest: state-of-the-art. R.C. Bittner, R. Felix. Strahlen- und Poliklinik, Charité, Campus
©ERS Journals Ltd 1998. Virchow-Klinikum, Medical Faculty, Hum-
ABSTRACT: To date, magnetic resonance (MR) is established as an imaging modal- boldt-University, Berlin, Germany.
ity in the diagnosis of chest diseases. Because of its excellent distinction of vessels and
Correspondence: R.C. Bittner
soft tissue, MR can be performed as the primary imaging procedure before computed Strahlen- und Poliklinik
tomography in patients with suspected vascular lesions, mediastinal masses, hilar Charité, Campus Virchow-Klinikum, HU
lesions, and pathological changes of the pleura and the chest wall. In these cases, MR Berlin
is able to provide all the necessary diagnostic information. In other patients, a limited Augustenburger Platz 1
number of MR images may be helpful in cases of equivocal or confusing CT or clini- D-13353 Berlin
cal findings. More detailed information can be obtained, using surface coils or special Germany
imaging sequences, i.e. high resolution MR images of the pleura or angiographic Fax: 49 30 450 57900
images of mediastinal and pulmonary vasculature.
Keywords: Contrast material
From a clinical viewpoint, the most important task for thoracic magnetic reso- magnetic resonance angiography
nance nowadays is the pretherapeutic evaluation of intrathoracic masses, the differ- magnetic resonance imaging
ential diagnosis of benign versus malignant lesions, and the accurate documentation mediastinum
of tumour extent in malignancies including three-dimensional-display to improve pleura
surgical or radiation planning. Future directions in thoracic magnetic resonance will thorax
be predominantly influenced by postprocessing approaches, specialized imaging tech-
niques, and magnetic resonance-guided interventional applications. Received: May 27 1997
Eur Respir J 1998; 11: 1392–1404. Accepted after revision January 20 1998

Since the early 1980s, magnetic resonance (MR) ima-
ging has been used for the evaluation of chest diseases
[1–43]. In the late 1980s, MR gained ground against the
established computed tomography (CT), but later lost this
due to the introduction of spiral CT [4, 6, 20, 34, 44–
52]. However, the main advantages of MR over CT are its
ability to perform multiplanar imaging and the excellent
soft tissue contrast. Another advantage is the depiction of
vascular structures without use of contrast material. In
recent years MR has been accepted beside the CT, and in
selected cases even as a primary diagnostic tool. Never-
theless, many applications of this imaging technique rem-
ain the subject of investigation; however, enough is known
today to suggest appropriate clinical indications for thora-
cic MR.
Technique
In contrast to other applications, the image quality of tho-
racic MR depends to a large extent on technical and exa-
miner conditions and experiences, because it is different
from MR examinations of other organs or regions. For most
chest MR exams, excellent electrocardiogram (ECG)-gat-
ing has to be achieved and prior patient's compliance is
required.
An advantage of MR over CT is the direct imaging in
the multiple planes of choice, which can provide informa-
tion that is not available on transaxial images. Structures
oriented longitudinally from cranial to caudal or oblique
can be imaged along their axis, such as the trachea, oeso-
phagus, superior vena cava, aorta or the brachial plexus.
Imaging in a second plane reduces the possibility of mis-
interpretation of findings as a result of partial volume
effects, i.e. in the area of the aorto-pulmonary window, the
subcarinal region, and in pathological processes near the
diaphragm or the lung apices. The most essential parame-
ters of the predominantly used imaging methods in thora-
cic MR [24, 41] shall be briefly addressed: basically, there
are two different imaging parameters, T1- and T2-weight-
ing, which lead to different signal intensities (SI) of diffe-
rent tissues on MR images. Thus, MR images are called
T1- or T2-weighted images, respectively. Fat as well as
MR contrast material are displayed with increased SI
(bright) in T1-weighted images. Water, fluid, and structu-
res with high water content present with low SI (dark) in
T1-weighted images. The T2-weighted image shows wa-
ter with increased SI and the same applies to changes with
a high water content, e.g. inflammatory changes, the majo-
rity of tumours, and nearly all pathological tissues (table
1).
For clinical use, T1-weighted images allow for excel-
lent soft tissue contrast-resolution and provide the best
spatial resolution for topographic evaluation. T2-weighted
images are necessary for the detection and assessment of
the extent of pathological changes with increased SI.
To date, in a routine chest examination (1.0–1.5 T magne-
tic field strength), a multi-slice (up to 32 sections), ECG-
gated, T1-weighted spin-echo sequence is generated within
3–7 min, in the transaxial and either coronal/sagittal or ob-
lique plane, respectively [19]. The slice thickness should be
3–8 mm, the field-of-view (FOV) may vary from 18–40 cm,
 MRI OF THE CHEST 1393

Table 1. – Characteristic signal intensity (SI) patterns of different tissues

Process T1-weighted T2-weighted Comment
Simple cyst Low High Homogeneous, sharply defined
Bronchogenic cyst Low-medium-high High-very high SI dependent on protein content, fluid level
possible
Seroma Low High Like cysts
Abscess Low-medium High Heterogeneous, ill-defined, often fluid level
Lipoma High Medium-high Sometimes brighter than normal fat
Neurinoma, schwannoma Low-medium Medium-high Heterogeneous, often centrally high SI on T2
Thymoma Low-medium Medium Homogeneous, sharply defined
Malignant lymphoma (Hodgkin's) Low-medium Low-high Heterogeneous, high SI of tumour interspersed
with low SI of fibrosis
Malignant lymphoma (non-Hodgkin's) Low-medium High Mostly homogeneous blurred margins
Bronchogenic carcinoma Low-medium Medium-high Mostly heterogeneous, blurred margins
Arteriovenous malformation Very low-signal void Medium-low Appearance dependent on flow
Obstructive atelectasis Low-medium High
Non-obstructive atelectasis Low-medium Low
Pulmonary secretions Low High
Alveolar proteinosis Medium-low Medium-low
Haemorrhage
Acute Low Low-medium Homogeneous
Subacute Medium-high High Heterogeneous fluid, fluid level
Chronic Low Medium-high Homogeneous with dark rim
Blood clot
Acute Medium Medium-high Magnetic
Chronic Medium Medium-low Susceptibility effects

and the spatial resolution is approximately 1–2 mm. In dis-
eases of the pleura and the chest wall or pulmonary proc-
esses near or in contact with the pleura, surface coils may
be used with reduction of the FOV down to 8 cm, allowing
for an excellent spatial resolution of 0.3 mm [11]. This may
be followed by a fast non-gated T2-weighted sequence in
one plane corresponding to the T1-weighted images, for
up to 26 images within 4–6 min [11, 41]. The T2-weighted
images display suspected increased SI and are helpful in
showing fluid collection, distinguishing tumour from fibro-
sis, or chest wall musculature, and identifying flow pheno-
mena, but the anatomical resolution is poor.
Several software programs have been introduced to eli-
minate flow, breathing and motion artifacts. In our experi-
ence, presaturation of the anterior chest wall provides the
most effective suppression of diminishing artifacts in lung

Fig. 1. – Magnetic resonance imaging of a severely ill patient with mediastinitis and extended soft tissue phlegmone spreading from an axillary abscess
(arrows). Coronal T1-weighted a) pre and b) postcontrast images show intense and extended contrast enhancement within the cervical, chest wall, neck,
and mediastinal soft tissue. Fast images, pre and postcontrast within 2.5 min.
1394 R.C. BITTNER, R. FELIX
and mediastinal MR exams [12, 19]. Postprocessing pro-
grams can significantly improve the image quality [4, 5, 35,
41].
The use of contrast material in thoracic MR is still a mat-
ter of controversy. Contrast-enhanced MR may be useful in
pulmonary and mediastinal masses to detect necrosis or
fibrosis [3–5, 7–13, 19, 20, 23, 28], and differential diag-
nosis may be improved [9–13]. Especially in pleural dis-
eases, in pulmonary processes adjacent to the pleura, and
in chest wall changes, contrast-enhanced T1-weighted MR
has been found to be advantageous because of its im-
proved ability to delineate lesions from normal structures
[8–11]. For imaging vascular structures and diagnosing
cardiac and vascular abnormalities, fast or dynamic ima-
ging techniques are most valuable. Fast dynamic (Cine-)
MR may show flowing blood with a very intense signal,
and ECG-gated, flow phenomena and cardiac function can
be demonstrated to great advantage [16, 23, 35, 41, 43].
Subsecond sequences allow for cine-images of the breath-
ing lung [21].
Rapid acquisition of spin-echo images may produce
good T1-weighted images during a breath-hold, so that
contrast-enhanced dynamic studies are feasible [1, 7, 23,
28, 29]. Thus, even emergency care patients can be exam-
ined within few minutes prior to and after contrast (fig. 1).
Respiratory gating significantly increases the examina-
tion time, is difficult to combine with ECG-gating and is
generally not used in clinical imaging [2, 19, 33, 36]. Tech-
nical progress has resulted in postprocessing procedures
with significant improvement of image quality without
prolongation of the patient's examination time and allow
for vivid preoperative three-dimensional (3D)-display of
thoracic masses [41].
Contraindications
There are no documented, lasting, harmful effects from
MR. In high field strength magnets, radio frequency-power
deposition may raise body core temperature, especially in
children [33, 41]. MR is absolutely contraindicated in
patients with cardiac pacemakers, even using low field
strength magnets. Because the magnetic field induces tor-
que on ferromagnetic implants like aneurysm or surgical
clips, patients with those materials should not be exam-
ined [24]. However, non-ferromagnetic implants are not
dangerous. Except for old Starr-Edwards type, all heart
valves can be imaged [24]. MR is also contraindicated in
the presence of metal objects within the eye or near the
spinal cord, cochlear implants, insulin pumps and neuro-
stimulators connected to the patient. Attention has to be
paid to shell fragments often seen in older patients. En-
capsulated, these are normally not dangerous, but within
the brain, chest or abdomen they may be dislocated and,
thus, cause occult bleeding.
Lung cancer (see also other chapters)
Since the advent of MR, numerous studies have been
performed to assess its value compared with CT in the
management of patients with lung cancer [2, 10, 12, 19,
22, 36]. In T-staging, central lung cancers >10 mm can be
detected even better than peripheral tumours [1, 2, 10, 29].
Fig. 2. – Pericardial invasion by tumour. T1-weighted parasagittal image,
widespread invasive bronchogenic carcinoma with disruption of the an-
terior pericardial fat layer by low signal intensity tumour (arrow), widen-
ing of the pericardial space due to malignant effusion (star). Note additional
tumorous chest wall invasion (arrowhead).
In assessing the tumour spread, there is good contrast
between malignant soft tissue and signal void vessels or
bronchi, intense mediastinal fat, and the mediastinal pleu-
ral margins. The unaltered pericardium is routinely dis-
played, identifiable as a line of hypointensity around the
heart and the base of the major vessels, itself surrounded
by a small fat plane [19, 33]. Obliteration of the mediasti-
nal fat planes, compression, encasement or involvement of
mediastinal vessels are better demonstrated by MR than
by CT [2, 19, 36] (fig. 2).
In our experience, aortic tumour invasion is suggested
if: 1) disruption of the periaortic fat layer of more than
90° of the circumference is present; 2) disruption of the
periaortic fat is visible more than 3 cm contiguously in
the cranio-caudal direction; and 3) focal disruption of the
periaortic fat is visible, including deformation of the lumen
[12, 19] (fig. 3). Furthermore, in selected cases focal en-
hancement of the aortic wall after contrast in the area of
adjacent tumour tissue may indicate malignant invasion.
Coronal or sagittal MR images are extremely valuable in
determining the precise extent of the tumour, especially in
relation to the carina and the aorto-pulmonary window, as
well as in relation to the pulmonary arteries and veins.
Involvement of the superior vena cava can be clearly ap-
preciated, and, in case of significant obstruction, typical
intraluminal thrombus with increased SI may be seen (fig.
4). Contrast-enhanced CT and MR are both accurate in
detecting hilar masses and adenopathy [2, 19, 47, 50]. In
cases of proximal bronchial obstruction associated with
postobstructive pneumonitis, MR may differentiate the
obstructing mass from adjacent lung on T2-weighted or
contrast-enhanced T1-weighted images [1, 2, 10, 12–14,
25, 29] (fig. 5). In such patients the consolidated lung typ-
ically exhibits a higher SI than the central hilar mass. MR
may even be successful in cases where differentiation is
not possible with dynamic enhanced CT scanning.
 MRI OF THE CHEST
Fig. 3. – Tumourous aortic invasion. a) Transverse and b) sagittal T1-weighted magnetic resonance images demonstrate broad tumour invasion of
the aortic wall by adjacent lung cancer. The perivascular fat layer has disappeared more than 180° of the circumference and more than 2 cm contiguously
in the course of the aorta. In addition, mediastinal tumour invasion causes obstruction of the left pulmonary artery, as well as of the anteriorly displaced
left main bronchus (arrows).
Fig. 4. – Thrombosis of the vena cava, T1-weighted coronal image.
Mediastinal lymph node metastasis (arrow) due to bronchogenic carci-
noma (not visible) causes occlusion of the superior vena cava with resul-
ting intense thrombus reaching up into the jugular vein (arrowheads).
In peripheral carcinomas, contact with the pleura may
suggest malignant involvement of the pleura, or of the
adjacent chest wall. A thin layer of extrapleural fat sepa-
1395
rating the tumour mass from the chest wall can almost
always be seen on high quality or high-resolution MR
images [8, 9, 11, 31]. In such cases contrast enhanced
T1-weighted MR is superior to CT in the demonstration
of malignant involvement. This is especially true in supe-
rior sulcus invasion, where CT is often equivocal, due to
partial volume effects (fig. 6). Sagittal or coronal plane
images often show the extent of chest wall invasion and
involvement of the subclavian artery or brachial plexus
better than either transaxial CT or MR images [8, 9]. In
patients who underwent radiation therapy for treatment of
carcinoma, recurrent tumour within radiation-induced fib-
rosis can be difficult to identify with CT. MR has the abi-
lity to differentiate tumour from fibrosis [2, 19]. By means
of high SI on T2-weighted images tumour relapse can be
distinguished from post-treatment fibrosis with low SI.
However, inflammatory reactions secondary to radiation
or from other causes may also show increased SI on T2-
weighted images.
Mediastinal masses
Generally, the relationship between a mediastinal mass
and adjacent vessels and a vascular compression or obstruc-
tion is better demonstrated by MR than by contrast-enhan-
ced CT, despite optimized contrast application with spiral
CT [2, 12, 19, 36, 46–48]. The differentiation of tumour
and mediastinal fat on T1-weighted MR is easy, since fat
shows higher SI on T1-weighted images. However, tumour
and mediastinal fat may be difficult or impossible to distin-
guish on T2-weighted images. The diagnosis of a media-
stinal mass or an enlarged lymph node does depend on
the spatial resolution and, even more important, on the soft
tissue contrast resolution. MR spatial resolution is only
slightly less than that of the fourth generation CT scanners,
while the soft tissue contrast resolution is far better than
1396 R.C. BITTNER, R. FELIX

Fig. 6. – Pancoast tumour. Left-sided sagittal T1-weighted magnetic

resonance image accurately displays tumour extension in Pancoast bron-
chogenic carcinoma with respect to superior sulcus structures. The tumour
has invaded the second rib (arrow), reaches the first rib and is already
adjacent to the lower fascicles of the brachial plexus, represented by linear
structures behind the cross-sectional displayed left subclavian artery
(arrowhead).

tense contrast enhancement of mediastinal lymph nodes in

MR suggests a limited number of differential diagnoses
including Castleman's disease, angioimmunoblastic lymp-
hadenopathy, as well as vascularized metastases, in particu-
lar, from renal cell carcinoma, papillary thyroid carcinoma,
and especially small-cell lung carcinoma [10, 12]. Further-
more, considerable contrast enhancement may be observed
in granulomatous disease such as tuberculosis or sarcoido-
sis, and in acquired immune deficiency syndrome (AIDS)-
related diseases, particularly Kaposi's sarcoma [3, 12, 26].
Since the presence of calcifications in a mass may be used
as evidence of benign disease, a significant disadvantage of
thoracic MR is that calcifications within a mediastinal mass
or a node are not sufficiently detectable (fig. 8). On the

Fig. 5. – Postobstructive atelectasis. a) Unenhanced transverse T1-weig-

hted magnetic resonance image in right central bronchogenic carcinoma,
causing atelectasis of the middle lobe. There is no reliable distinction of
tumour and parenchymal changes. b) Corresponding T2-weighted image
with relatively low signal of centrally located tumour (arrow) and easily
appreciable postobstructive atelectasis with high signal intensity (arrow-
heads).

that of CT [12, 19, 36]. The accuracy of both MR and CT

in diagnosing mediastinal lymph node metastases, i.e. in
patients with lung cancer, is comparable, since lymph node
size is the sole criterion for determining tumour involve-
ment [2, 19, 22, 36] (fig. 7). It is recommended that short
axis of subcarinal lymph nodes should not exceed 11 mm,
10 mm for right tracheobronchial, right paraoesophageal,
low paratracheal, and aorto-pulmonal, and 7–8 mm for all Fig. 7. – Aorto-pulmonary lymph nodes. a) Thin-section computed tomo-
other nodal groups, according to the American Thoracic graphy with good contrast bolus, suspicion of small mediastinal lymph
Society system [22, 54]. Neither relaxation times, nor SI, nodes, partially calcified, in the level of the aorto-pulmonary window
(arrow). b) Coronal T1-weighted magnetic resonance image clearly de-
nor the degree of contrast enhancement are reliable indica- monstrates three small lymph nodes without partial volume effects
tors of malignant involvement [10, 12, 22]. However, in- (arrowhead).
 MRI OF THE CHEST
other hand, in some cases, particularly when the aorto-pul-
monary window or the subcarinal space are involved, MR
is able to demonstrate lymph nodes better than CT because
of its ability to image variable planes (fig. 7).
Fig. 8. – Calcified mediastinal lymph node. a) computed tomography
shows large calcification in a pericarinal lymph node in a patient with
multiple mediastinal and hilar metastases due to breast carcinoma. b)
Corresponding transverse T1-weighted magnetic resonance image allows
only suggestion of calcification due to signal-void in this region (arrow).
1397
Regarding evaluation of the thymus, MR is comparable
with CT [12, 38]. On T2-weighted images, visualization
of the normal gland is more difficult than on T1-weighted
images [33]. However, mediastinal thymolipoma can be
diagnosed by signal characteristics due to the fat content.
In patients with myasthenia gravis, no specific changes
of the thymus can be seen [33]. In patients with malig-
nant diseases, very often a thymic rebound phenomenon is
visible on sagittal mediastinal T1-weighted images. On
MR images, mediastinal solid thymomas characteristical-
ly present with sharply defined margins, triangular con-
tours, with medium SI on T2-weighted images and none
to slight enhancement on T1-weighted images after con-
trast [12].
In most cases of lymphoma, the CT density of fibrous
and active lymphomatous tissue shows no significant dif-
ference, even after sufficient intravenous contrast applica-
tion. Lymphomatous cells with larger amount of water
show a relatively lower proportion of protein. Conversely,
fibrosis contain much less water and a high proportion of
protein. This leads to the markedly different T2 appear-
ance of lymphoma and fibrosis in MR images. In nodular-
sclerosing Hodgkin's disease (HD), a substantial amount
of sclerosis can be seen interspersed with malignant cells.
Conversely, in diffuse non-Hodgkin's lymphoma (NHL),
many more malignant cells and fewer interspersed fibrous
tissues may be found. Thus, characteristic signal patterns
for lymphomas on T2-weighted MR images can be de-
fined as follows:
1) A homogeneous or homogeneously fine-nodular hy-
perintense pattern is characteristic for untreated nonscle-
rosing lymphoma [12, 19]. In T1-weighted images those
masses show a homogeneous low SI similar to muscle.
2) A mixed hyper-/hypointensity pattern is often seen in
untreated nodular-sclerosing HD, where low-signal areas
Fig. 9. – Hodgkin's disease, nodular-sclerosing type. Transverse T2-
weighted image shows inhomogeneous signal intensity with ring-shaped
low-signal fibrosis (arrowheads); typical appearance of nodular-sclerosing
Hodgkin's lymphoma.
1398 R.C. BITTNER, R. FELIX

within small "islands" in residual masses previously consi-

Fig. 10. – Residual mass in malignant non-Hodgkin's lymphoma. a)
Chest radiograph; and b) computed tomography (CT) show extended
soft tissue in the anterior mediastinum after chemotherapy and radiation
of malignant B-cell lymphoma. Differentiation of fibrosis and tumour
relapse is not possible. c) Corresponding (to CT) transverse T2-weighted
magnetic resonance (MR) image displays fibrosis nearly signal-void,
interspersed fat with higher signal intensity (SI) corresponds with low
attenuation values in CT and high SI in T1-weighted MR (not shown).
High SI is due to flow phenomena in the anterior azygos vein and fluid
in the upper retroaortic pericardial recess.
dered inactive suggests tumour recurrence. Nevertheless,
cautious interpretations should be made within the first 6
months following therapy as regards inflammation and ne-
crosis [19].
Cystic or fluid-filled masses, or necrotic changes can be
detected by means of low SI on T1- and high SI on T2-
weighted images, even when CT numbers suggest solid
masses [19]. Detection of cystic or fluid-filled masses can
also be made on the basis of no or minimal enhancement
after contrast [10]. Thus, MR can be successfully used to
diagnose a wide range of lesions, including bronchogenic
cysts, pericardial cysts, thymic cysts, colloid cysts within
goiters, dermoid cysts, cystic hygromas, and even medias-
tinal pseudocysts, especially complex cysts that do not
appear fluid-filled on CT.
The typical MR appearance of both benign and malig-
nant neurogenic tumours includes slightly greater SI than
muscle on T1-weighted images, and moderately to mark-
edly increased SI on T2-weighted images. Compared with
CT, MR has several distinct advantages for imaging par-
aspinal neurogenic tumours [12, 39]. Intraforaminal/spinal
extension can be clearly assessed, as well as associated
spinal cord pathology [39]. Coronal or oblique sagittal
images may help to assess the relationship to the sympa-
thetic nerve. The detection of tumour infiltration beyond
the parietal pleura suggests malignancy of a neurogenic
mass [8, 10].
Midsagittal MR images can reliably demonstrate cra-
nio-caudal tumour extent in patients with oesophageal
carcinoma in presurgical evaluation, and suspected tum-
our invasion of adjacent structures can be confirmed or
excluded with great sensitivity [19] (fig. 11).

represent sclerotic tumour regions, nodular configuration

may also be found [12, 19] (fig. 9). In addition, this pat-
tern is seen during the response phase of most lympho-
mas under treatment, representing residual tumour and
necrosis or inflammation beside fibrosis. Mixed
fibro-fatty masses are easy to recognize on T1-weighted
images, since the fat portions will exhibit high SI.
3) Hypointense patterns are characteristic for inactive
residual fibrotic masses following successful therapy for
lymphoma and may be seen in up to 88% of patients [33]
(fig. 10). These lesions also show low SI on T1-weighted
images.
Thus, monitoring of SI in lymphomas by MR can con-
tribute to therapeutic management: a decrease in mass size
and a corresponding SI decrease presume a favourable
response. A decrease in size with persistent homogeneous/
heterogeneous hyperintense patterns suggest a partial res-
Fig. 11. – Oesophageal carcinoma. Midsagittal T1-weighted image
ponse. Marked size regression with small residual masses demonstrates direct cranio-caudal extent of oesophageal tumour (arrow-
of heterogeneous/homogeneous high SI strongly suggests heads). Note the unaffected normal oesophagus with slightly widened
inappropriate response of the tumour parts. SI increase lumen below the tumorous changes.
 MRI OF THE CHEST
Hilar abnormalities
The diagnosis of a hilar mass on CT images requires
differentiation of normal or vascular tissue from abnormal
soft tissue. In some locations, these differentiations can be
made on anatomical grounds alone, but in other areas
mass and vessels may be difficult to distinguish unless a
precise time-controlled or large bolus of contrast medium
is given. However, despite the introduction of spiral CT, in
a significant number of cases the differentiation of mass
and vessels after contrast is inadequate [2, 36, 47]. This
problem can be avoided by using MR. Since only the walls
of pulmonary arteries and veins are visible on regular T1-
weighted MR images, hilar masses are easy to detect. In
particular, sagittal images provide excellent spatial resolu-
tion of the hilar architecture and can detect normal size or
enlarged lymph nodes [2, 19, 36].
In patients with malignant disease, contrast-enhanced
CT and MR are both quite accurate in detecting hilar
masses or node enlargement, with a sensitivity approach-
ing 100% [2, 19, 36, 47, 50]. On the other hand, specificity
in the detection of lymph nodes harbouring metastases,
such as in the mediastinum, is rather poor using either
method [22, 54]. Unfortunately, MR is unable to detect
nodal calcification. On the other hand, in patients with
poor vascular opacification in CT, MR allows a more con-
fident diagnosis of a normal or abnormal hilum. In gen-
eral, bronchi are more accurately evaluated with CT. MR
may be advantageous in showing significant hilar or medi-
astinal vascular invasion contiguous with the hilar mass,
and in precisely displaying its extent [2, 19, 36].
Hilar masses, i.e. bronchogenic carcinoma, and adjacent
pulmonary obstructive collapse can be distinguished with
MR [2, 14, 19, 25, 36]. Generally, tumour tissue presents
with higher SI than distal lung tissue in T2-weighted ima-
ges. Conversely, in tumour obstruction, distal consolidated
lung parenchyma appears more intense than the tumour.
Parenchymal lung diseases
On spin-echo images in normal subjects, little signal
is obtained from the lung parenchyma, mainly because
of the small number of protons, magnetic susceptibility
effects and the very short T2-relaxation time of lung pa-
renchyma [41]. Fine linear structures extending laterally
from the hilum into the lung represent the walls of vessels
and/or bronchi. Increased SI in the parenchyma itself,
often visible in the posterior area of both lungs, may rep-
resent condensation of the lung parenchyma in a supine
position and an increase of intravascular blood. This sig-
nal is increased considerably after intravenous application
of contrast material. Because of the signal void of normal
lung parenchyma, abnormalities (nodules, masses, paren-
chymal changes) are easily detectable within the lungs.
However, because of the invisibility of the usual structures
like segmental bronchi and vessels, it may be difficult to
determine the precise location of a pathological process.
Nevertheless, the normal fissures are reliably displayed on
sagittal images [3]. It has been shown that routine MR
with 8 mm sections from the apex to the adrenals is able to
successfully depict pulmonary nodules greater than 5 mm,
given adequate patient compliance [2, 19, 36]. Altogether,
CT is superior to MR in the diagnosis of parenchymal
1399
lung disease. Lung nodules greater than 1 cm are shown in
both MR and CT, but the morphological characteristics of
a nodule, such as edge definition, spiculation, and associ-
ated pleural tail are better defined with CT [4, 46, 47, 50].
Unfortunately, the presence of calcification in a nodule,
important in distinguishing benign and malignant disease,
is only reliably visible in thin-section or high-resolu-
tion CT (fig. 8). Overall, MR is less sensitive than CT
in detecting lung nodules <1 cm in diameter, mostly due
to respiratory motion during the MR study. However, in
centrally-located nodules, MR may be superior to CT in
distinguishing these processes from vessels [2, 19]. In pul-
monary consolidation, fluid replaces air within the lung
parenchyma. MR may be able to characterize the fluid or
the cause of consolidation and, thus, may be able to distin-
guish diseases. Nonobstructive atelectasis shows a poor
signal in T2-weighted images and only minimal enhance-
ment after contrast [10, 25].
New fast sequences allow for better evaluation of inter-
stitial lung changes, as well as cine-images of the breath-
ing lung with the moving diaphragm and chest wall during
controlled inspiration and expiration in patients with bul-
lous emphysema and possible surgical therapy [1, 21, 29,
41]. Although spiral CT provides excellent 3D-visualiza-
tion of pulmonary lesions, vascular abnormalities within
the lungs like arteriovenous malformations in Osler dis-
ease are displayed by MR angiography with great accu-
racy [4, 17, 51].
Pleura and chest wall
Pleural and chest wall abnormalities can occur in a
number of benign and malignant diseases, e.g. tuberculo-
sis, asbestosis, malignant mesothelioma, lung cancer, or
metastatic disease. Due to the excellent soft tissue con-
trast, improved spatial resolution, and imaging in various
planes, MR has been proven to be especially suitable for
the evaluation of pleural and chest wall abnormalities [8–
11, 15, 31, 37].
Although some correlations have been found both in
vitro and in vivo between MR SI and pleural fluid compo-
sition, MR cannot differentiate between various aetio-
logies of pleural effusions, but can distinguish free and
loculated effusions, as well as identify coexistent under-
lying lung disease [9, 11, 31]. An accurate demonstration
of pleural fluid composition is difficult with MR, but
blood in the pleural space is distinguishable from other
fluids, except for acute bleeding [41].
CT diagnosis of pleural disease is based on the finding
of pleural effusion and pleural thickening, but the distinc-
tion of pleural and lung parenchymal abnormalities from
pleural effusion may be difficult [9]. Although CT studies
could demonstrate some progress in distinguishing malig-
nant from benign pleural diseases, the infiltration of the
chest wall, together with osseous destruction, still remains
the only reliable sign of malignancy [9]. Recent results of
MR examinations in patients with different pleural and
chest wall diseases could demonstrate that MR can be
extremely advantageous in these cases, improving both
quality of evaluation and differential diagnosis [8, 9, 11,
15]. MR detects pleural diseases with high sensitivity,
comparable with CT, and has a reliable potential in differ-
ential diagnosis of pleural changes, based on the improved
1400 R.C. BITTNER, R. FELIX

Fig. 12. – High-resolution anatomical magnetic resonance (MR) visualization of the brachial plexus. a) Oblique coronal T1-weighted MR image with
typical appearance of the band-shaped fascicles of the brachial plexus (arrows) extending superiorly to the (left) subclavian artery (A) (V: subclavian
vein). b) Cross-sectional sagittal view (according to the white line in (a)); behind the clavicle (C) the low-signal fascicles of the brachial plexus (arrows)
are easily appreciable within the fat superior to the subclavian artery (A) and vein (V). c) Tumour invasion of the brachial plexus. Oblique coronal T1-
weighted MR image shows low-signal tumour tissue of the metastatic breast cancer invading the fascicles of the plexus (arrows) and presumably
involving the adjacent left subclavian artery (A).

distinction of pleural changes and adjacent structures after
contrast enhancement [9, 11]. On the basis of morpholog-
ical patterns such as nodular changes, thickening >10 mm,
mediastinal or circumferential pleural involvement, or ple-
ural extension through the entire hemithorax, malignant
disease is suggested [9, 11]. The most specific pattern for
malignancy is the infiltration of the chest wall and/or the
diaphragm. MR, especially when contrast-enhanced, has
demonstrated its superiority over CT in the diagnosis of
diaphragm and chest wall invasion in patients with differ-
ent pulmonary masses, although spiral CT has regained
some ground [45]. Particularly in tumours of the superior
sulcus or the lung base, images in the coronal or sagittal
plane can be used to demonstrate the relationship of the
tumour within the lung apex to adjacent structures such as
the brachial plexus, and the subclavian artery and vein [2,
8–11, 15, 37] (fig. 6). The normal and pathological appe-
arance of the brachial plexus is well demonstrated by T1-
weighted MR images, especially regarding displacement
and encasement/infiltration of the subclavian artery by
tumour (fig. 12). Invasion of the diaphragm can be visua-
lized by contrast enhancement of the regions involved [8–
11] (fig. 13). The invasion of the thin layer of extrapleural
fat by tumour tissue is a reliable criterion for early inva-
sion of the chest wall [11, 15]. This fat layer is more diffi-
cult to visualize using CT.
While on unenhanced T1-weighted images tumour tis-
sue can be easily distinguished from fat, the T2-weighted

Fig. 13. – Tumour invasion of the diaphragm. a) Unenhanced and b) contrast-enhanced coronal T1-weighted magnetic resonance images in malignant
pleural mesothelioma show intense band-shaped enhancement of the lateral diaphragm (arrows), representing tumour invasion (surgically confirmed).
 MRI OF THE CHEST 1401

Fig. 14. – Primary pleural liposarcoma. a) Transverse T2-weighted magnetic resonance image displays pleural and fissural (arrow) changes with high
signal intensity. No reliable distinction of pleural effusion and tumourous changes. b) Corresponding contrast-enhanced T1-weighted image. Excellent
differentiation of enhancing irregularly defined parietal tumour nodules (black arrows), non-enhancing effusion (star), slightly thickened visceral pleura
(arrowheads), and fissural tumour involvement (white arrow).

and contrast-enhanced T1-weighted image allow reliable
distinction of tumour and muscle. Contrast-enhanced T1-
weighted images are superior to T2-weighted images bec-
ause of their improved spatial resolution with comparable
lesion contrast [8–11]. Therefore, and because of the
absolute increase in signal within the pleural space in the
presence of pleural changes on T2-weighted images, the
method of choice in pleural MR should be unenhanced
and contrast-enhanced T1-weighted sequences (fig. 14).
Improved imaging of pleural pathology and chest wall
invasion is achieved by using surface coils. High-resolu-
tion MR studies of the pleural space and the inner chest
wall suggest a reliable approach in the differential diagno-
sis of pleural diseases, and in the early detection of malig-
nant chest wall invasion [9, 11] (fig. 15).
Vascular lesions, heart and pericardium
MR imaging has proven valuable in the diagnosis of a
number of vascular changes involving the mediastinum
[12, 16, 18–20, 34, 35, 43]. Aortic aneurysms are easily
identified with MR because of the signal void usually
associated with flowing blood. Organized thrombi within
the lumen of aneurysms usually appear as areas of inter-
mediate to increased SI on T1-weighted images. T1-
weighted MR may be limited in differentiating between
mediastinal fat and subacute or chronic haemorrhage be-
cause of a leaking aneurysm, owing to potential overlap in
their SI parameters [33–35]. MR and spiral CT are used
equally in screening for aortic dissection [47, 53]. ECG-
gated T1-weighted spin-echo sequences, best acquired in
transverse and coronal/sagittal planes, can demonstrate in-
timal flap in patients with dissection, with an accuracy
equalling or exceeding that of CT [34, 35].
In patients with superior vena cava obstruction, the area
of narrowing or obstruction can be demonstrated [2, 12,
19, 33, 36]. Furthermore, the evidence of slow flow is
seen, and increased SI within the lumen of the superior
vena cava suggests the diagnosis of thrombosis [19] (fig.
4). However, a fresh thrombus may appear similar to flow-
ing blood and may be missed on unenhanced or nonangio-
graphic MR images [33, 41]. Contrast-enhanced MR is
able to differentiate thrombus from intravascular tumour
[10, 12]. On MR angiographic images, central emboli are
displayed as low intensity defects within the high signal of
flowing blood [6, 20] (fig. 16). Nevertheless, in pulmo-
nary artery embolism, to date, spiral CT with bolus-appli-
cation of contrast material is the method of choice [4, 20,
35, 44, 47, 49, 53]. In patients with tumour involvement of
the main pulmonary arteries or veins, MR can better dem-
onstrate the tumorous extension, compared with CT [2,
12, 19, 36, 43].
In the evaluation of the heart and the pericardium, the
advantages of MR include larger FOV than ultrasound and
far better soft-tissue contrast resolution than ultrasound
and CT. While cine-MR is believed to be the most sensi-
tive method for detecting pericardial effusion and intracav-
itary masses, myocardial and pericardial masses are best
imaged with spin-echo techniques [2, 12, 16, 18, 23, 24,
1402 R.C. BITTNER, R. FELIX

Fig. 15. – Malignant invasion of the chest wall. a) Computed tomography

shows pleura-based tumour mass with broad contact to the intercostal
soft tissue and presumable invasion (arrow). b) Conventional coronal T1-
weighted magnetic resonance (MR) image demonstrates broad tumour
contact with the lateral chest wall. c) High-resolution sagittal T1-weighted
MR image excellently reveals focal invasion of the inner chest wall with
disruption of the peripleural fat, involvement of the innermost intercostal
muscle and the intercostal fat (arrow). Surgery confirmed focal chest wall
invasion by malignant fibrous histiocytoma.

28, 30, 43]. MR is particularly sensitive for the presence of

pericardial fluid, especially intrapericardial haemorrhage
[2, 19, 43]. When combined pericardial and cardiac in-
flammatory or tumorous involvement is suspected, MR is
usually superior to CT and ultrasound [2, 12, 19] (fig. 2).
Summary
Compared with CT, the main advantages of MR are the
ability to conduct multiplanar imaging, the excellent soft
tissue contrast, the depiction of vascular structures with-
out the use of contrast material, and the ability to produce
detailed angiographic images of mediastinal and pulmo-
nary vasculature, already comparable with digital sub-
straction angiography. The the main disadvantage of MR
compared with CT in clinical routine is the longer exami-
nation time, in addition to the insufficient demonstration
of calcifications. MR is used as a primary imaging moda-
lity in patients with suspected vascular lesions, mediasti-
nal masses, hilar lesions, and pathological changes of the
pleura and the chest wall. In many other patients, MR is
Fig. 16. – Pulmonary artery thrombosis. a) T1-weighted transverse image
shows low-signal thrombus adjacent to the anterior vascular wall (arrow)
with compression of the lumen of the superior vena cava (arrowhead). b)
Parasagittal magnetic resonance-angiogram demonstrates nearly signal-
void thrombus (star), high-signal true lumen of the enlarged right pulmo-
nary artery (arrow), and medium-signal of the compressed superior vena
cava (arrowhead).
used as a problem-solving modality in answering specific
questions raised by the CT study or clinical findings. De-
tailed information can be achieved by high-resolution MR
regarding pleural and chest wall changes, especially after
contrast application. MR is able to distinguish different
tissues, fluids, or pathological processes within the chest.
Furthermore, qualitative and quantitative data regarding
blood flow, volume and blood pressure within vessels and
the heart, as well as diaphragm function can be obtained.
Postprocessing algorithms, which are being increasingly
used, can eliminate motion, flow, breathing and other arti-
facts, and significantly improve MR image quality.
 MRI OF THE CHEST
Future outlook
About 15 years after the introduction of magnetic re- 13.
sonance imaging into clinical routine, there are emerg-
ing applications expanding magnetic resonance from a
purely diagnostic modality to its use in guiding invasive 14.
procedures throughout the body. Dedicated interventio-
nal magnetic resonance-systems with open magnets are
being developed, allowing direct access to patients dur-
ing procedures such as magnetic resonance-endoscopy, 15.
stereotactic biopsy, radiofrequency tumour ablation, or
laser discectomy. Magnetic resonance-guided-biopsy and
laser-induced interstitial thermotherapy has already been 16.
established in liver, breast and neck abnormalities. In pre-
therapeutic planning three-dimensional-image reconstruc-
tion and "virtual reality" processing of simulated vascular, 17.
intestinal and bronchial endoscopies, as well as simulated
surgical interventions, are being developed. Real-time ma-
gnetic resonance techniques allow for reliable functional
imaging comparable with ultrasound, and tissue tempera- 18.
tures can be noninvasively measured and also used for
thermal imaging. 19.
References
20.
1. Alsop DC, Hatabu H, Bonnet M, Listerud J, Gefter W.
Multi-slice, breathhold imaging of the lung with submilli-
second echo times. MRM 1995; 33: 678–682.
2. Armstrong P, Reznek RH, Phillips RR. Diagnostic imag- 21.
ing of lung cancer. In: Spiro SG, ed. Carcinoma of the
Lung. Eur Respir Mon 1995; 1: 137–187.
3. Auffermann W, Hieckel HG, Bittner RC, et al. Magnet-
resonanz-Tomographie der Lunge bei pulmonaler Sarkei- 22.
dose. Pneumologie 1992; 46: 277–281.
4. Beier J, Liebig T, Bittner RC, et al. Computergraphische
Methoden zur dreidimensionalen Darstellung intrapul-
monaler Raumforderungen aus CT- und MR-Bildern. 23.
Pneumologie 1996; 50: 672–678.
5. Beier J, Bittner RC, Wust P, Fleck E, Kaiser D, Felix R.
3-D-Darstellung von thorakalen Tumoren aus MR-Bil-
dern. Fortschr Roentgenstr 1996; 164: 75–78. 24.
6. Bergin CJ, Hauschildt J, Rios G, Belezzuoli EV, Huynh
T, Channick RN. Accuracy of MR angiography compared
with radionuclide scannning in identifying the cause of
pulmonary arterial hypertension. AJR 1997; 168: 1549– 25.
1555.
7. Berthezène Y, Vexler V, Clément O, Mühler A, Moseley
ME, Brasch RC. Contrast-enhanced MR imaging of the 26.
lung: assessments of ventilation and perfusion. Radiology
1992; 183: 667–672.
8. Bittner RC, Kaiser D, Loddenkemper R, Felix R. Wertig-
keit der Magnetresonanztomographie bei Pancoast- und 27.
anderen Tumoren mit Infiltration der Thoraxwand. Pneu-
mologie 1990; 44: 62–69.
9. Bittner RC, Schoerner W, Loddenkemper C, et al. Pleura- 28.
erkrankungen in der Magnetresonanztomographie (MRT).
Pneumologie 1992; 46: 612–620. 29.
10. Bittner RC, Schoerner W, Auffermann W, et al. Contrast
enhanced magnetic resonance imaging in chest patholo-
gies. Am Rev Respir Dis 1992; 145: A196.
11. Bittner RC, Schnoy N, Schoenfeld N, et al. Hochauflö-
sende Magnetresonanztomographie (HR-MR) von Pleura 30.
und Thoraxwand: Normalbefund und pathologische Ve-
ränderungen. Fortschr Roentgenstr 1995; 162: 296–303.
12. Bittner RC, Schoenfeld N, Loddenkemper R, Kaiser D, 31.
Dorow P, Felix R. Improved differential diagnosis with
1403
magnetic resonance imaging (MRI) in mediastinal patho-
logies. Am J Respir Crit Care Med 1995; 151: A846.
Bockisch A, Daniel G, Harvey RC, Davis MA, Kabalka
GW. MRI contrast enhancement of the lung using a Gd-
DTPA aerosol. Radiologica Diagnostica 1993; 34: 103–105.
Bourgouin PM, McLoud TC, Fitzgibbon JF, et al. Differen-
tiation of bronchogenic carcinoma from postobstructive
pneumonitis by magnetic resonance imaging: histopatho-
logic correlation. J Thorac Imaging 1991; 6: 22–27.
Carlsen SE, Bergin CJ, Hoppe RT. MR imaging to detect
chest wall and pleural involvement in patients with lym-
phoma: effect on radiation therapy planning. AJR 1993;
160: 1191–1195.
Chin AJ, Fogel MA. Noninvasive Imaging of Congenital
Heart Disease. Before and After Surgical Reconstruction.
New York, Futura, 1994.
Dinsmore BJ, Gefter WB, Hatabu H, Kressel HY. Pulmo-
nary arteriovenous malformations: diagnosis by gradient-
refocused MR imaging. J Comput Assist Tomogr 1990;
14: 918–923.
Dueriockx AJ. Coronary MR angiography. MRI Clin
North Am 1996; 4: 361–418.
Felix R, Bittner RC. Wertigkeit der Magnetresonanztom-
ographie (MRT) in der Pneumologie. Internist 1993; 34:
1038–1043.
Gefter WB, Hatabu H, Holland GA, Gupta KB, Henschke
CI, Palevsky HI. Pulmonary thromboembolism: recent
developments in diagnosis with CT and MR imaging.
Radiology 1995; 197: 561–574.
Gierada DS, Curtin JJ, Erickson SJ, Prost RW, Strandt JA,
Goodman LR. Diaphragmatic motion: fast gradient-
recalled-echo MR imaging in healthy subjects. Radiology
1995; 194: 879–884.
Glazer GM, Orringer MB, Chenevert TL, et al. Mediasti-
nal lymph nodes: relaxation time/pathologic correlation
and implications in staging of lung cancer with MR imag-
ing. Radiology 1988; 168: 429–431.
Hartnell GG, Cerel A, Kamalesh M, et al. Detection of
myocardial ischemia: value of combined myocardial per-
fusion and cineangiographic MR imaging. AJR 1994;
163: 1061–1067.
Hartnell GG, Spence L, Hughes LA, Cohen MC, Saouef
R, Buff B. Safety of MR imaging in patients who have re-
tained metallic materials after cardiac surgery. AJR 1997;
168: 1157–1159.
Herold CJ, Kuhlmann JE, Zerhouni EA. Pulmonary atel-
ectasis: signal pattern with MR imaging. Radiology 1991;
178: 710–715.
Khalil AM, Carette MF, Cadranel JL, Mayaud CM,
Akoun GM, Bigot JM. Magnetic resonance imaging find-
ings in pulmonary Kaposi's sarcoma: a series of 10 cases.
Eur Respir J 1994; 7: 1285–1289.
Leung DA, Debatin JF, Wildermuth S, et al. Intravascular
MR tracking catheter: preliminary experimental evalua-
tion. AJR 1995; 164: 1265–1270.
Lombardi M. Ultrafast MR methods detect and diagnose
myocardial infarction. Diagn Imag Int 1994; 11: 56–59.
Lutterbey G, Gieseke J, Sommer T, Keller E, Kuhl C,
Schild H. Ein neuer Ansatz in der Magnetresonanztom-
ographie der Lunge mit einer ultrakurzen Turbo-Spin-
Echo-Sequenz (UTSE). Fortschr Roentgenstr 1996; 164:
388–393.
Masui T, Finck S, Higgins CB. Constructive pericardi-
tis and restrictive cardiomyopathy: evaluation with MR
imaging. Radiology 1992; 182: 369–373.
Müller NL. Imaging of the pleura. Radiology 1993; 186:
297–309.
1404 R.C. BITTNER, R. FELIX
32. Mumtaz H, Hall-Craggs MA, Wotherspoon A. Laser
therapy for breast cancer: MR imaging and histopatho-
logical correlation. Radiology 1996; 200: 651–658.
33. Naidich DP, Zerhouni EA, Siegelman SS. Computed
Tomography and Magnetic Resonance of the Thorax, 2nd
edition. New York, Raven Press, 1991.
34. Nienaber CA, von Kodolitsch Y, Nicolas V, et al. The
diagnosis of thoracic aortic dissection by noninvasive
imaging procedures. N Engl J Med 1993; 328: 1–9.
35. Prince MR, Narasimhan DL, Jacoby WT, et al. Three-
dimensional Gadolinium-enhanced MR angiography of
the thoracic aorta. AJR 1996; 166: 1387–1397.
36. Quint LE, Francis IR, Wahl RL, Gross BH, Glazer GM.
Preoperative staging of non-small-cell carcinoma of the
lung: imaging methods. AJR 1995; 164: 1349–1359.
37. Rapoport S, Blair DN, McCarthy SM, Desser TS, Ham-
mers LW, Sostman HD. Brachial plexus: correlation of
MR imaging with CT and pathologic findings. Radiology
1988; 167: 161–165.
38. Sakai F, Sone S, Kiyono K, et al. MR imaging of thy-
moma: radiologic-pathologic correlation. AJR 1992; 158:
751–756.
39. Sakai F, Sone S, Kiyono K, et al. Intrathoracic neuro-
genic tumors: MR-pathologic correlation. AJR 1992; 159:
279–283.
40. Silverman JM, Julien PJ, Herfkens RJ, Pelc NJ. Quantita-
tive differential pulmonary perfusion: MR imaging versus
radionuclide lung scanning. Radiology 1993; 189: 699–
701.
41. Stichnoth FA. MR-Tomographie. Technische Grundlagen
und klinische Aspekte, 2nd edition. Berlin, Blackwell
Wissenschaft, 1994.
42. Vogl TJ, Muller PK, Hammerstingl R. Malignant liver tu-
mors treated with MR imaging-guided laser-induced
thermotherapy: technique and prospective results. Radiol-
ogy 1995; 196: 257–265.
43. Wetter DR, McKinnon GC, Debatin JF, von Schulthess
GK. Cardiac echo-planar MR imaging: comparison of
single- and multiple-shot techniques. Radiology 1995;
194: 765–770.
44. Bergin CJ, Rios G, King MA, Belezzuoli EV, Luna J,
Auger WR. Accuracy of high-resolution CT in identi-
fying chronic pulmonary thromboembolic disease. AJR
1996; 166: 1371–1377.
45. Brink JA, Heiken JH, Semenkovich J, Teefey SA, McCle-
nnan BL, Sagel SS. Abnormalities of the diaphragm and
adjacent structures: findings on multiplanar spiral CT
scans. AJR 1994; 163: 307–310.
46. Buckley JA, Scott WW Jr, Siegelman SS, et al. Pulmo-
nary nodules: effect of increased data sampling on detec-
tion with spiral CT and confidence in diagnosis. Radiology
1995; 196: 395–400.
47. Heiken JP, Brink JA, Vannier MW. Spiral (helical) CT.
Radiology 1993; 189: 647–656.
48. Kauczor HU, Wolcke B, Fischer B, Mildenberger P,
Lorenz J, Thelen M. Three-dimensional helical CT of the
tracheobronchial tree: evaluation of imaging protocols
and assessment of suspected stenoses with bronchoscopic
correlation. AJR 1996; 167: 419–424.
49. Kuzo RS, Goodman LR. CT evaluation of pulmonary
embolism: technique and interpretation. AJR 1997; 169:
959–965.
50. Paranjpe DV, Bergin CJ. Spiral CT of the lungs: optimal
technique and resolution compared with conventional CT.
AJR 1994; 162: 561–567.
51. Remy J, Remy-Jardin M, Giraud F, Wattinne L. Angioar-
chitecture of pulmonary arteriovenous malformations:
clinical utility of three-dimensional helical CT. Radiology
1994; 191: 657–664.
52. Vining DJ, Liu K, Choplin RH, Haponik EF. Virtual
bronchoscopy: relationships of virtual reality endobron-
chial simulations to actual bronchoscopic findings. Chest
1996; 109: 549–553.
53. Beier J, Diebold T, Liebig T, Biamino G, Fleck E, Felix
R. Virtual endoscopy in the assessment of aortic lesions.
AJR 1997; 168: S178.
54. Kiyono K, Sone S, Sakai F, et al. The number and size of
normal mediastinal lymph nodes: a postmortem study.
AJR 1988; 150: 771–776.